Mineral resource quantitative prediction method and system based on knowledge embedding

Abstract

The invention provides a mineral resource quantitative prediction method and system based on knowledge embedding, and the method comprises the steps: constructing a prospecting conceptual model according to a metallogenic geologic body, a metallogenic structure system and a mineralization feature mark related to a target mining area; according to the prospecting conceptual model and prospecting geological knowledge of the target mining area, obtaining a plurality of feature variables, interpolating discrete and discontinuous feature variable sampling points of the target mining area into continuous data through a spatial interpolation mode, and generating a prediction feature map layer; determining the weight of each characteristic variable; generating a prediction model according to the weight of each feature variable and a preset ore deposit information feature depth network; a plurality of samples are generated based on the prediction feature map layer, one sample corresponds to one unit grid, and the samples comprise a positive sample, a negative sample and a to-be-predicted sample; and inputting the plurality of samples into the prediction model to obtain a mineral resource prediction result of the target mining area. According to the technical scheme, the prediction accuracy of the mineral yield can be improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of mineral exploration, and in particular to a method and system for quantitative prediction of mineral resources based on knowledge embedding. Background Art

[0002] With the development of science and technology and the advancement of mineral exploration, geoscience data (such as geological, geophysical, geochemical, remote sensing and other data) has continued to grow, providing a solid data foundation for prospecting and exploration. At the same time, with the support of advanced computer science and artificial intelligence technology, digital earth science has developed rapidly, and various quantitative mathematical models and calculation methods have emerged in an endless stream. Intelligent mineral prediction systems with machine learning and deep learning methods as the core have advantages in processing large-scale data and complex pattern recognition, and can support the extraction and expression of deep-level characteristics of mineralization information, establish nonlinear associations from mineralization elements to ore bodies, and have made some good progress in the prediction and evaluation of mineral resources.

[0003] With the deepening of understanding of geological processes and the development and improvement of modern mineralization theories and technical methods, the prediction and evaluation of mineral resources has developed from traditional qualitative prediction research to quantitative evaluation based on multiple models; from simple similarity analogy to complex multi-source geological data abnormal information mining and integration, from the prediction and evaluation of a single mineral species to the comprehensive evaluation of multiple mineral species; from the qualitative prediction and evaluation of ore deposit output to the comprehensive evaluation and quantitative prediction of mineral resources of different scales and scales (ore belts, ore fields, ore deposits, etc.). The prediction and evaluation of mineral resources also has a systematic, scientific, visual and dynamic trend. Machine learning, especially deep learning models, have excellent processing capabilities for complex and nonlinear problems, making the intelligent mineral prediction system a powerful tool for analyzing complex geological mineralization processes and conducting mineralization prediction reasoning in the process of mineralization prediction.

[0004] However, the above prediction methods in the prior art still have limitations, especially the inability to accurately predict mineral output. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method and system for quantitative prediction of mineral resources based on knowledge embedding, which can improve the prediction accuracy of mineral output.

[0006] The embodiment of the present invention provides a method for quantitative prediction of mineral resources based on knowledge embedding, comprising:

[0007] Construct a prospecting conceptual model based on the mineralization geological bodies, mineralization structural systems and mineralization characteristics associated with the target mining area;

[0008] According to the prospecting concept model and the prospecting geological knowledge of the target mining area, a plurality of characteristic variables are obtained, and discrete and discontinuous characteristic variable sampling points of the target mining area are interpolated into continuous data by means of spatial interpolation to generate a prediction characteristic layer;

[0009] Determining the weight of each of the characteristic variables;

[0010] Generate a prediction model based on the weights of each of the characteristic variables and a pre-set ore deposit information characteristic depth network;

[0011] Based on the prediction feature layer, a plurality of samples are generated, wherein one sample corresponds to one unit grid, a plurality of the unit grids constitute the prediction feature layer, and the samples include positive samples, negative samples and samples to be predicted;

[0012] A plurality of the samples are input into the prediction model to obtain the mineral resource prediction result of the target mining area.

[0013] Optionally, the step of determining the weight of each of the feature variables includes:

[0014] Determine a first score and a second score for each characteristic variable;

[0015] Compare the first feature vector with the highest first score with other feature variables, and determine a first score vector based on the relative importance between the first feature vector and other feature variables;

[0016] Comparing the second feature vector with the lowest second score with other feature variables, and determining a second score vector based on the relative importance between the second feature vector and the other feature variables;

[0017] According to each feature variable, a first scoring vector value of each feature variable in the first scoring vector, and a second scoring vector value of each feature variable in the second scoring vector, a BWM knowledge embedding model is used to determine a weight of each feature variable.

[0018] Optionally, the step of determining the first score and the second score of each characteristic variable includes:

[0019] Extract key information elements from each characteristic variable and determine the key information expression of each characteristic variable;

[0020] Input the key information expression of each feature variable into the preset knowledge graph model to obtain multiple key nodes corresponding to each key information expression;

[0021] Using the knowledge graph model, multiple key nodes of the key information expression are mapped to obtain multiple vector representations corresponding to each key information expression;

[0022] The multiple vector representations corresponding to each key information expression and each key information expression are fused to form a fused variable corresponding to each feature variable;

[0023] The fused variables corresponding to each feature variable are input into the pre-trained neural network, and under different confidence conditions, the first confidence and the second confidence of each fused variable are respectively obtained, and the first confidence is used as the first score, and the second confidence is used as the second score.

[0024] Optionally, the step of generating a prediction model according to the weights of the characteristic variables and a preset ore deposit information characteristic depth network includes:

[0025] The weight of the characteristic variable is input into the mineral deposit information characteristic depth network loss function to determine the loss value of the mineral deposit information characteristic depth network, wherein the loss value is determined based on the following formula:

[0026]

[0027] Based on the loss value, back propagation is performed using a gradient descent method to determine the parameter gradient of the mineral deposit information feature depth network;

[0028] Based on the parameter gradient, using a parameter optimization algorithm, the weight parameters between different layers of the ore deposit information feature depth network are updated to generate the prediction model;

[0029] Where N is the number of samples, C is the number of categories; y ij is a sample X i The true label of the j-th category in; is a sample X i The predicted probability of the jth category in w j is the weight.

[0030] Optionally, the sample is a positive sample, and the positive sample is used to characterize a mineral point;

[0031] The step of generating a plurality of samples based on the predicted feature layer comprises at least one of the following:

[0032] Gridding the predicted feature layer to generate a plurality of cells; generating a buffer zone with the mineral points in the target mining area as the center according to the spatial correlation of mineralization, and taking all samples in the buffer zone as initial samples, wherein the range of the buffer zone is determined based on the area of ​​the target mining area and the number of mineral points in the target mining area;

[0033] Taking the cell where the initial sample point is located as a base point, any one cell from a plurality of neighboring cells is selected as an extended sample; performing difference processing on the extended sample and the initial sample to generate a plurality of the samples;

[0034] The initial samples are labeled to generate initial samples with first label information and sample data with second label information are obtained; a preset supervision model is trained using the data with the first label information to obtain a learning model; the sample data with the second label information is input into the learning model to generate sample data with pseudo label information; the initial samples with the first label information and the sample data with the pseudo label information are input into the learning model to generate a plurality of the samples.

[0035] Optionally, the sample is a negative sample, and the negative sample is used to characterize a non-mineral point;

[0036] The step of generating a plurality of samples based on the predicted feature layer comprises at least one of the following:

[0037] Gridding the predicted feature layer to generate a plurality of cells, wherein one cell corresponds to one eigenvalue; calculating the statistics of each eigenvalue, and according to the statistics of each eigenvalue and the preset statistics, taking the samples corresponding to the cells outside the preset statistics as negative samples;

[0038] Based on a nonlinear dimensionality reduction method, the high-dimensional data in the prediction feature layer is mapped to a low-dimensional space, the abnormal area is determined, and the samples corresponding to the cells in the abnormal area are taken as negative samples.

[0039] Optionally, the step of inputting the prediction feature layer and the plurality of samples into the prediction model to obtain the mineral resource prediction result of the target mining area includes:

[0040] Based on the positive sample, the negative sample and the sample to be predicted, as well as the first label corresponding to the positive sample, the second label corresponding to the negative sample and the third label corresponding to the sample to be predicted, and based on the prediction model, determine the mineralization probability of each unit grid;

[0041] Based on the mineralization probability of each unit grid, the cumulative frequency method is used to determine the classification of the target area, and the cumulative result is used as the mineral resource prediction result of the target mining area, wherein the mineral resource prediction result of the target mining area includes the delineation of the predicted target area and the mineralization probability within the predicted target area.

[0042] Optionally, before inputting the prediction feature layer and the plurality of samples into the prediction model to obtain the mineral resource prediction result of the target mining area, the prediction method further comprises:

[0043] The prediction model is deconstructed and the parameters of the prediction model are adjusted using permutation importance, partial dependency plots, and model-independent local explanation algorithms.

[0044] Accordingly, an embodiment of the present invention further provides a mineral resource quantitative prediction system based on knowledge embedding, comprising:

[0045] A construction unit is configured to construct a prospecting conceptual model based on the mineralization geological bodies, mineralization structural systems and mineralization characteristic signs related to the target mining area;

[0046] The first generating unit is configured to obtain a plurality of characteristic variables according to the prospecting conceptual model and the prospecting geological knowledge of the target mining area, and interpolate the discrete and discontinuous characteristic variable sampling points of the target mining area into continuous data by means of spatial interpolation to generate a prediction characteristic layer;

[0047] A processing unit is configured to determine the weight of each of the characteristic variables; and based on the prediction characteristic layer, generate a plurality of samples, wherein one sample corresponds to one unit grid, a plurality of the unit grids constitute the prediction characteristic layer, and the samples include positive samples, negative samples and samples to be predicted; and input the plurality of the samples into a prediction model to obtain a mineral resource prediction result of the target mining area;

[0048] The second generating unit is configured to generate the prediction model according to the weights of the respective characteristic variables and a preset mineral deposit information characteristic depth network.

[0049] Correspondingly, an embodiment of the present invention also provides an electronic device, comprising: at least one memory and at least one processor, the memory storing one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the quantitative prediction method of mineral resources based on knowledge embedding as described in any of the aforementioned embodiments.

[0050] Compared with the prior art, the technical solution of the embodiment of the present invention has the following advantages:

[0051] In the quantitative prediction method of mineral resources based on knowledge embedding provided by the embodiment of the present invention, by constructing a prospecting concept model related to the target mining area, a prediction feature layer can be generated based on the prospecting concept model and the prospecting geological knowledge of the target mining area, and then multiple samples can be generated based on the prediction feature layer; at the same time, a prediction model is generated based on the weights of each characteristic variable and a pre-set ore deposit information feature depth network, so that the prediction result of the mineral resources in the target mining area can be determined by inputting multiple samples into the prediction model. Since the prospecting concept model takes into account the ore-forming geological body, ore-forming structural system and ore-forming characteristic signs, the entire prediction process can be deconstructed from a multi-dimensional and multi-level model from the global and local, feature and sample, and at the same time, the geological knowledge is added to the loss value calculation of the ore deposit information feature depth network in the form of weights in the characteristic variables, and a knowledge-data-jointly driven intelligent mineral prediction system is obtained, which effectively improves the accuracy, stability and reliability of the prediction model, thereby improving the prediction accuracy of mineral output, which helps to promote the exploration and development of hidden minerals and is of great significance for achieving a new round of prospecting breakthroughs. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a flow chart of a method for quantitative prediction of mineral resources based on knowledge embedding in an embodiment of the present invention;

[0053] Figure 2 is a flow chart of determining the weights of each of the characteristic variables in an embodiment of the present invention;

[0054] Figure 3 Schematic diagram of the process of obtaining the weights of each characteristic variable in an embodiment of the present invention;

[0055] Figure 4 is a schematic diagram of a generation process of a prediction model in an embodiment of the present invention;

[0056] Figure 5 is a schematic diagram of a positive sample acquisition process in the first embodiment of the present invention;

[0057] Figure 6 is a schematic diagram of a positive sample acquisition process in the second embodiment of the present invention;

[0058] Figure 7 is a schematic diagram of an output of a prediction model in an embodiment of the present invention;

[0059] Figure 8 It is a structural schematic diagram of a mineral resource quantitative prediction system based on knowledge embedding in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] As can be seen from the background technology, the prediction accuracy of mineral resources obtained by existing machine (or deep) learning models is still limited. This is because:

[0061] The construction of machine (or deep) learning models is mainly based on data-driven, relatively lacking the guidance of geological knowledge, and the "black box" effect is inevitable in the intelligent prediction process, which makes the current data-driven artificial intelligence methods face two important challenges: purely data-driven machine (or deep) learning models lack the guidance of geological knowledge, and it is difficult to fully correspond to the characteristics of the ore deposit genesis model based solely on spatial data characteristics, and there is a lack of cognitive consistency; at the same time, machine (or deep) learning algorithms can mine the deep-level characteristics of mineralization information, but because of their nested nonlinear network structure and abstract expression, they have highly opaque black box properties, and the interpretability of the model is relatively weak, which makes each stage have different sources of uncertainty, and will be propagated to the next stage, making the results of mineral prediction uncertain.

[0062] In order to solve the above technical problems, an embodiment of the present invention provides a method for quantitative prediction of mineral resources based on knowledge embedding, comprising: constructing a prospecting concept model according to the mineralization geological body, mineralization structural system and mineralization characteristic signs related to the target mining area; obtaining multiple characteristic variables according to the prospecting concept model and the prospecting geological knowledge of the target mining area, and interpolating the discrete and discontinuous characteristic variable sampling points of the target mining area into continuous data through spatial interpolation to generate a prediction feature layer; determining the weight of each of the characteristic variables; generating a prediction model according to the weight of each of the characteristic variables and a pre-set ore deposit information feature depth network; generating multiple samples based on the prediction feature layer, wherein one sample corresponds to one unit grid, multiple unit grids constitute the prediction feature layer, and the samples include positive samples, negative samples and samples to be predicted; inputting multiple samples into the prediction model to obtain the mineral resource prediction result of the target mining area.

[0063] In the quantitative prediction method of mineral resources based on knowledge embedding provided by the embodiment of the present invention, by constructing a prospecting concept model related to the target mining area, a prediction feature layer can be generated based on the prospecting concept model and the prospecting geological knowledge of the target mining area, and then multiple samples can be generated based on the prediction feature layer; at the same time, a prediction model is generated based on the weights of each characteristic variable and a pre-set ore deposit information feature depth network, so that the prediction result of the mineral resources in the target mining area can be determined by inputting the prediction feature layer and multiple samples into the prediction model. Since the prospecting concept model takes into account the mineralization geological body, the mineralization structural system and the characteristic signs of the mineralization, the entire prediction process can be deconstructed from a multi-dimensional and multi-level model from the global and local, the characteristics and the samples, and an intelligent mineral prediction system driven by knowledge-data joint is obtained, which effectively improves the accuracy, stability and reliability of the prediction model, and then improves the prediction accuracy of the mineral output, which helps to promote the exploration and development of hidden minerals and is of great significance for achieving a new round of prospecting breakthroughs.

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the present invention is described below with reference to the accompanying drawings.

[0065] See also Figure 1 A flowchart of a method for quantitative prediction of mineral resources based on knowledge embedding in an embodiment of the present invention is shown in FIG. Figure 1 As shown, the following prediction steps can be performed:

[0066] Step A: Construct a prospecting conceptual model based on the mineralization geological bodies, mineralization structural systems and mineralization characteristic signs related to the target mining area.

[0067] The target mining area (as a non-limiting example, the target mining area may refer to the Korne ore field) may refer to a mineral resource area that has been determined to have economic mining value after exploration and evaluation.

[0068] A mineralizing geological body refers to an aggregate of mineral resources formed through a series of geological processes under specific geological environments and conditions. It usually includes ore bodies, veins, and deposits.

[0069] The mineralization structural system refers to the geological structural system where mineral resources are formed and accumulated in a specific geological environment.

[0070] The characteristic signs of mineralization refer to the characteristics and signs shown in the process of mineral resource formation. These characteristics can help geologists and mineral explorers identify and understand the mineralization process.

[0071] In other words, the mineralization geological body, mineralization structural system and characteristic signs of mineralization are all used to characterize the degree of mineralization in the target mining area. Based on the above factors, the prospecting concept model can include different exploration perspectives, improve the accuracy of the prospecting concept model, and provide a solid basis for intelligent prediction.

[0072] In this embodiment, the steps of constructing a prospecting conceptual model may include:

[0073] A pre-trained target detection model is used to respectively identify the first structural element in the ore-forming geological body, the second structural element in the ore-forming structural system and the third structural element in the characteristic mark of the ore-forming process; according to the first structural element, a first triplet relationship between the ore-forming geological body and the initial prospecting concept model is constructed, according to the second structural element, a second triplet relationship between the ore-forming structural system and the initial prospecting concept model is constructed, and according to the third structural element, a third triplet relationship between the characteristic mark of the ore-forming process and the initial prospecting concept model is constructed; the first triplet relationship is randomly initialized to obtain a first embedding vector, the second triplet relationship is randomly initialized to obtain a second embedding vector, and the third triplet relationship is randomly initialized to obtain a third embedding vector; the ComplEx model is used to train the first embedding vector, the second embedding vector and the third embedding vector to respectively determine the first knowledge graph corresponding to the ore-forming geological body, the second knowledge graph corresponding to the ore-forming structural system, and the third knowledge graph corresponding to the characteristic mark of the ore-forming process; the first knowledge graph, the second knowledge graph and the third knowledge graph are fused using the attention mechanism to obtain the prospecting concept model.

[0074] By constructing a prospecting concept model based on the knowledge graph, the accuracy of the prospecting concept model can be improved, which is conducive to improving the identification and analysis capabilities of mineral resources.

[0075] In this embodiment, as shown in Table 1, the prospecting conceptual model can be expressed in the following forms:

[0076] Table 1 Schematic diagram of prospecting concept model

[0077]

[0078] Step B, based on the prospecting conceptual model and the prospecting geological knowledge of the target mining area, obtain multiple characteristic variables, and interpolate the discrete and discontinuous characteristic variable sampling points of the target mining area into continuous data through spatial interpolation to generate a prediction feature layer.

[0079] Specifically, prospecting geological knowledge mainly involves the distribution, genesis, exploration and development of mineral resources, while the prospecting conceptual model contains relevant mineral information in the target mining area. In this way, multiple characteristic variables can be selected based on the prospecting conceptual model and prospecting geological knowledge. As shown in Table 2, the characteristic variables can include geological, remote sensing and geochemical data. In order to reflect the spatial correlation of the data, the discrete and discontinuous characteristic variable sampling points of the target mining area are interpolated into continuous data through spatial interpolation processing (as a non-limiting example, spatial interpolation processing can include inverse distance weighted method and Kriging method) to generate a predicted feature layer, so that the predicted feature layer can more realistically represent the actual mineral distribution in the target mining area.

[0080] Table 2 Schematic table of characteristic variables

[0081]

[0082] Step C, determining the weight of each of the characteristic variables.

[0083] As shown in Table 2, there are multiple feature variables, and the types of the multiple feature variables are not exactly the same. Thus, the weights of the various feature variables are also different. Therefore, when making predictions, the importance of different feature variables is also different, which results in differences in the emphasis of the prediction model finally formed.

[0084] In this embodiment, see Figure 2 , the step of determining the weights of each of the feature variables may include:

[0085] Step C1, determining a first score and a second score for each characteristic variable.

[0086] Specifically, based on the characteristic variables included in Table 2, the first score and the second score of each characteristic variable may be predetermined, wherein the first score and the second score may refer to the scores of the characteristic variables under different evaluation systems.

[0087] In this embodiment, the step of determining the first score and the second score of each feature variable includes: extracting key information elements in each feature variable to determine the key information expression of each feature variable; inputting the key information expression of each feature variable into a preset knowledge graph model to obtain multiple key nodes corresponding to each key information expression; using the knowledge graph model, mapping the multiple key nodes of the key information expression to obtain multiple vector representations corresponding to each key information expression; fusing the multiple vector representations corresponding to each key information expression and each key information expression to form a fused variable corresponding to each feature variable; inputting the fused variable corresponding to each feature variable into a pre-trained neural network, and under different confidence conditions, respectively obtaining the first confidence and the second confidence of each fused variable, and using the first confidence as the first score, and using the second confidence as the second score.

[0088] By determining the first score and the second score of each feature variable based on the knowledge graph model, the knowledge graph model can fully understand the importance of each feature variable to obtain more reliable and more accurate first score and second score.

[0089] Step C2: compare the first feature vector with the highest first score with other feature variables, and determine a first score vector based on the relative importance between the first feature vector and other feature variables.

[0090] Specifically, based on step C1, the feature variable with the highest first score can be obtained from multiple feature variables as the first feature vector (it can be understood that the number of first feature vectors is at least 1), so that the first feature vector can be used as a reference benchmark, and according to the first scores of other feature variables, the first score vector of each feature variable in this case can be determined.

[0091] Step C3: compare the second feature vector with the lowest second score with other feature variables, and determine a second score vector based on the relative importance between the second feature vector and other feature variables.

[0092] Specifically, based on step C1, the feature variable with the lowest second score can be obtained from multiple feature variables as the second feature vector (it can be understood that the number of second feature vectors is at least 1), so that the second feature vector can be used as a reference benchmark, and according to the second scores of other feature variables, the second score vector of each feature variable in this case can be determined.

[0093] Step C4, according to each feature variable, the first scoring vector value of each feature variable in the first scoring vector, and the second scoring vector value of each feature variable in the second scoring vector, the BWM knowledge embedding model is used to determine the weight of each feature variable.

[0094] Specifically, the BWM (Best-Worst Method) knowledge embedding model integrates the knowledge embedding technology of the best-worst method. After determining each feature variable and the first scoring vector and the second scoring vector of each feature variable through steps C1 to C3, the weight of each feature variable can be determined based on the processing logic of the BWM knowledge embedding model itself.

[0095] As an alternative example, see Figure 3 The schematic diagram of the process of obtaining the weights of each characteristic variable in this embodiment is shown in FIG. Figure 3 The neutron graph (a) represents the relative importance of the values ​​assigned to each characteristic variable, where the smaller the value, the lower the relative importance of the characteristic variable (i.e., 9 is the most important and 1 is the least important). Figure 3 The sub-graph (b) is used to characterize the weights of each characteristic variable after steps C1 to C4, and it can be seen from the sub-graph (b) that different characteristic variables have different weights.

[0096] For example, the spectrum of albite has a weight of 0.16, while the weight of B is 0.025.

[0097] It should be noted that Figure 3 The schematic diagram of the process of obtaining the weights of the characteristic variables is only an example and should not be construed as a limitation to the present invention.

[0098] Step D, generating a prediction model according to the weights of each of the characteristic variables and a pre-set ore deposit information characteristic depth network.

[0099] Specifically, all characteristic variables can characterize the distribution of minerals in the target mining area, and the weights of the characteristic variables reflect the importance of the characteristic variables. Based on the weights of each characteristic variable, the pre-set mineral deposit information feature deep network can be pre-trained to generate the corresponding prediction model.

[0100] See also Figure 4 The schematic diagram of the generation process of the prediction model in the embodiment of the present invention is shown in FIG. Figure 4 As shown, based on the prospecting geological knowledge and in conjunction with the BWM knowledge embedding model, the weights of the feature weight variables can be determined.

[0101] Then, the weights of the characteristic variables are input into the loss function of the mineral deposit information characteristic depth network to determine the loss value of the mineral deposit information characteristic depth network, and then based on the loss value, the gradient descent method is used for back propagation to determine the parameter gradient of the mineral deposit information characteristic depth network, and finally based on the parameter gradient, the parameter optimization algorithm is used to update the weight parameters between different layers of the mineral deposit information characteristic depth network to generate the prediction model (as a non-limiting example, by updating the weight parameters between the same layers, all the characteristic variables are input into the mineral deposit information characteristic depth network again, and forward propagation is performed to determine whether one or more verification evaluation indicators of the mean square error and absolute error of the prediction results meet the requirements, and the prediction model can be determined).

[0102] Among them, based on the parameter gradient, the parameter optimization algorithm is used to update the weight parameters between different layers of the ore deposit information feature depth network. Please refer to the description in the existing scheme. The focus of this scheme is: how to determine the loss value based on the weight of the feature variable.

[0103] In this way, based on the weights of characteristic variables and the loss function of the deep network of deposit information characteristics, a data-driven intelligent prediction model can be obtained to enhance the model generalization ability and result reliability.

[0104] In this embodiment, the loss value is determined based on the following formula:

[0105]

[0106] Where N is the number of samples, C is the number of categories (as a non-limiting example, if it is a binary classification task, then C = 2); y ij is a sample X i The true label of the jth category in (as a non-limiting example, if the sample belongs to category j, then y ij =1, otherwise y ij =0); is a sample X i The predicted probability of the jth category in w j is the weight, that is, the weight of the feature variable obtained by step C.

[0107] In this embodiment, the deep network of mineral deposit information features can be a network architecture with a fully connected layer as the main body, including: an input layer, which is composed of multiple first neurons; an output layer, which is composed of multiple second neurons; and multiple hidden layers for connecting the input layer and the output layer, and each hidden layer is interconnected.

[0108] Step E: generating a plurality of samples based on the predicted feature layer, wherein one sample corresponds to one unit grid, and a plurality of the unit grids constitute the predicted feature layer.

[0109] The samples include positive samples, negative samples and samples to be predicted.

[0110] Specifically, after the prediction feature layer is cropped or gridded, multiple layers are merged into multi-channel data to adapt to the input of the model. After the grids where the positive and negative sample labels are located are determined according to the positive and negative sample label setting method, the grid at this time includes the label data and the feature variable data. The grid where the positive sample label is located is regarded as a positive sample, the grid where the negative sample label is located is a negative sample, and the remaining grids have unknown labels and are samples to be predicted.

[0111] Among them, positive samples are used to characterize mineral points, and negative samples are used to characterize non-mineral points. The characterization content of the samples to be tested is unknown and needs to be determined in this scheme.

[0112] In other words, the correctness of label setting has a great impact on the performance and prediction accuracy of the intelligent prediction algorithm, and the prediction feature layer can represent the geological outline of the area, so the prediction feature layer can be processed to generate positive and negative samples.

[0113] It should be noted that when generating multiple samples, the following conditions should be met:

[0114] Representativeness: Positive and negative samples should be representative and reflect the geological characteristics and mineral distribution in the study area.

[0115] Independence: Positive and negative samples should be independent of each other to avoid the impact of correlation between samples on model training.

[0116] Balance: The number of positive and negative samples should be balanced to avoid the impact of an imbalance in the ratio of positive and negative samples on model training.

[0117] Accuracy: The labels of positive and negative samples should be accurate to avoid label errors affecting model training.

[0118] In this embodiment, the sample is a positive sample, and the positive sample is used to characterize a mineral point. In this case, at least one of the following methods may be used to generate multiple samples.

[0119] Method 1: See Figure 5In the schematic diagram of the process of obtaining positive samples in the first embodiment shown, the predicted feature layer is gridded to generate multiple cells; according to the spatial correlation of mineralization, a buffer zone is generated with the mineral point of the target mining area as the center, and all samples in the buffer zone are used as initial samples, wherein the range of the buffer zone is determined based on the area of ​​the target mining area and the number of mineral points in the target mining area (as a non-limiting example, if the area of ​​the target mining area is large and there are many mineral points, the buffer zone distance can be appropriately reduced and the number of buffer zones can be appropriately increased); the cell where the initial sample point is located is taken as the base point, and any one cell among multiple neighboring cells is selected as an extended sample; the extended sample and the initial sample are differenced to generate multiple samples.

[0120] Using method 1 can avoid information loss and reduce the possibility of noise generation, while ensuring the quality of the generated positive samples, which is consistent with geological cognition.

[0121] Method 2: See Figure 6 The schematic diagram of the positive sample acquisition process in the second embodiment shown in the figure labels the initial sample to generate an initial sample with a first label information, and obtains sample data with a second label information; trains a preset supervision model with the first label information to obtain a learning model; inputs the sample data with the second label information into the learning model to generate sample data with pseudo label information; inputs the initial sample with the first label information and the sample data with the pseudo label information into the learning model to generate a plurality of the samples.

[0122] By adopting method 2, a large number of positive samples can be generated, thereby improving the prediction accuracy of the prediction model.

[0123] In this embodiment, the sample is a negative sample, and the negative sample is used to characterize a non-mining point. In this case, at least one of the following methods can be used to generate multiple samples:

[0124] Method 1: Grid the predicted feature layer to generate multiple cells, where one cell corresponds to one eigenvalue; calculate the statistics of each eigenvalue, and based on the statistics of each eigenvalue and the preset statistics, take the samples corresponding to the cells outside the preset statistics as negative samples.

[0125] In short, by analyzing the distribution of each feature in the data, the thresholds of mine-induced anomalies and mine-induced anomalies are determined, and the areas within the threshold range can be used as negative samples.

[0126] Specifically, statistical analysis is performed on each characteristic value, and its mean value, standard deviation and other statistical quantities are calculated to set a reasonable threshold range. For example, for lithological characteristics, non-ore-bearing rock bodies or rock bodies far away from ore-bearing rock bodies must be selected; for the concentration of geochemical elements, the abnormal concentration threshold can be determined by analyzing the frequency distribution of the data.

[0127] Method 2: Based on a nonlinear dimensionality reduction method, the high-dimensional data in the prediction feature layer is mapped to a low-dimensional space, the abnormal area is determined, and the samples corresponding to the cells in the abnormal area are used as negative samples.

[0128] In short, negative sample selection based on nonlinear dimensionality reduction methods, that is, using nonlinear dimensionality reduction methods (such as PCA, t-SNE) to map high-dimensional data to low-dimensional space and identify non-abnormal areas.

[0129] Specifically, PCA or t-SNE is used to reduce the dimension of the data to a two-dimensional or three-dimensional space. By visualizing the points in these low-dimensional spaces, the distribution of the samples can be intuitively identified. Sample points far away from the positive samples are selected as negative samples.

[0130] Based on the above methods 1 and 2, representative and accurate negative samples can be carefully selected, thereby improving the prediction accuracy of the prediction model.

[0131] It should be noted that after determining the positive samples and negative samples, the positive samples and negative samples can also be labeled to better distinguish between the positive samples and negative samples (as a non-limiting example, the label of the positive sample is +1 and the label of the negative sample is -1); then the positive samples and negative samples can be combined to form different data sets, and all data sets are input into the data-driven intelligent prediction model, and the prediction results are evaluated to select the best setting method to form the optimal input data set.

[0132] Step F, inputting the plurality of samples into the prediction model to obtain the mineral resource prediction result of the target mining area.

[0133] Specifically, by adopting the above steps, a prediction model, a prediction feature layer and multiple positive and negative sample labels are formed at the same time, and the prediction feature layer can be used as a feature, and multiple positive and negative sample labels can be used as training sample labels, so as to predict the mineralization probability of samples with unknown labels based on each unit. In this way, the mineralization probability prediction results of the comprehensive study area are combined, and the target area classification is determined using the cumulative frequency method, thereby realizing the delineation of the prediction target area, so as to further determine the mineral resources of the target mining area, that is, to obtain the prediction results.

[0134] For further information, see Figure 7The output schematic diagram of the prediction model shown, based on the positive samples, negative samples and samples to be predicted, as well as the first label corresponding to the positive samples, the second label corresponding to the negative samples and the third label corresponding to the samples to be predicted, based on the prediction model, the mineralization probability of each unit grid is determined; based on the mineralization probability of each unit grid, the target area classification is determined using the cumulative frequency method, and the cumulative result is used as the mineral resource prediction result of the target mining area, wherein the mineral resource prediction result of the target mining area includes the delineation of the prediction target area and the mineralization probability within the prediction target area.

[0135] It should be noted that this scheme adopts a prediction model based on weights and outputs prediction results based on samples obtained from the prediction feature layer. The reason is that the existing deep learning loss function does not take the weight of the variables into account, which makes the accuracy of the final output prediction results low, and the mineral deposits as point data are difficult to reflect the spatial correlation of the data.

[0136] Therefore, through spatial interpolation processing, the mineral deposit is transformed into a prediction feature layer, and the subsequent input data is also an image. In addition, there are no training samples and prediction samples at the beginning. In order to ensure the prediction effect of the model, it is important to set the correct training samples. It is necessary to interpolate the existing discrete sampling points of the target mining area (the sampling point data is the coordinates and the corresponding feature variable values) into an image to obtain the feature variable layer. The prediction feature layer is equivalent to the feature, and there is no label at this time.

[0137] The optimal input data set obtained by setting the positive and negative sample labels is equivalent to finding the location of the training samples, and then finding the locations of these training samples on the prediction feature layer. Through image cropping, that is, gridding, the grid where the positive sample label is located is regarded as a positive sample, and the grid where the negative sample label is located is regarded as a negative sample. At this time, the grid contains the value and label of the feature. A grid can be regarded as a training sample, and some grids without labels are samples to be predicted.

[0138] In this way, through the above method, on the one hand, the accuracy of the prediction model is improved; on the other hand, the accuracy of the sample is also improved, so that the prediction result is more accurate.

[0139] In this embodiment, in order to further improve the prediction accuracy of the prediction result of the mineral resources of the target mining area, before inputting the prediction feature layer and the plurality of samples into the prediction model to obtain the prediction result of the mineral resources of the target mining area, the prediction method further includes:

[0140] The prediction model is deconstructed and the parameters of the prediction model are adjusted using permutation importance, partial dependency plots, and model-independent local explanation algorithms.

[0141] Specifically, by deconstructing the prediction model, potential mineralization knowledge can be discovered, and re-knowledge iteration can be formed to adjust the model parameters of the data-driven prediction model to obtain optimized mineral resource prediction results.

[0142] Moreover, due to the lack of reliability of unexplainable mineral prediction models, and because machine learning models are similar to black boxes and cannot control the internal operation of the model, we can only try different parameters. Therefore, this study introduces permutation importance, partial dependence plot (PDP) and model-independent local explanation (LIME) algorithms.

[0143] Among them, permutation importance is an effective method to measure feature importance that is independent of the prediction model. By changing the eigenvalue of a feature, the importance of the value is proved. The greater the change, the greater the importance.

[0144] The specific steps are as follows: 1) After the prediction model is trained, select a feature; 2) Set a random number for the feature and calculate a new prediction result; 3) Compare the new and old prediction results to obtain the impact of the feature on the model prediction. The global feature weights obtained by applying this scheme can be compared with the prior feature weights calculated by geological knowledge during the model construction phase, thereby analyzing the similarities and differences between the feature and mineralization relationship in the data-driven model and the knowledge-driven model.

[0145] The basic idea of ​​the partial dependency diagram is to change the focused feature of a machine learning model, keep other features unchanged, and observe the impact of this feature on the prediction results. At the same time, PDP can illustrate the relationship between feature variables and prediction results.

[0146] When the partial dependence diagram is applied to this scheme, the influence trend of each feature on the mineralization prediction probability is different. Taking the two features of Na2O+K2O geochemical element combination and annular structure as examples for analysis, it can reflect the fluctuation of mineralization probability with the change of feature values.

[0147] The local explanation algorithm is a model-independent local interpretability algorithm that can truly reflect the behavior of the classifier when predicting samples. It is applied to this solution to randomly select positive and negative samples from labeled samples to perform feature analysis on a single sample, and compare the intelligent prediction results with the manual annotation results to achieve cross-validation.

[0148] In short, through interpretable methods, we can deconstruct the model prediction results, quantitatively evaluate the importance of prediction indicators, form a loss function that integrates geological knowledge, adjust model parameters, optimize the prediction effect, realize the updating and iteration of mineral deposit knowledge, and form a repeatable, quantifiable, and controllable closed loop of intelligent mineral prospecting and prediction research.

[0149] The embodiment of the present invention further provides a system corresponding to the method for quantitative prediction of mineral resources based on knowledge embedding, which can be referred to as "quantitative prediction system 100 for mineral resources". Figure 8 The schematic diagram of the structure of a mineral resource quantitative prediction system based on knowledge embedding in an embodiment of the present invention is shown in FIG. Figure 8 As shown, the mineral resources quantitative prediction system 100 may include:

[0150] The construction unit 110 is configured to construct a prospecting conceptual model according to the ore-forming geological bodies, ore-forming structural systems and ore-forming characteristic marks related to the target mining area;

[0151] The first generating unit 120 is configured to obtain a plurality of characteristic variables according to the prospecting concept model and the prospecting geological knowledge of the target mining area, and interpolate the discrete and discontinuous characteristic variable sampling points of the target mining area into continuous data by means of spatial interpolation to generate a prediction characteristic layer;

[0152] The processing unit 130 is configured to determine the weight of each of the characteristic variables; and generate a plurality of samples based on the prediction characteristic layer, wherein one sample corresponds to one unit grid, a plurality of the unit grids constitute the prediction characteristic layer, and the samples include positive samples, negative samples and samples to be predicted; and input the plurality of the samples into the prediction model to obtain the mineral resource prediction result of the target mining area;

[0153] The second generating unit 140 is configured to generate the prediction model according to the weights of the respective characteristic variables and a preset ore deposit information characteristic depth network.

[0154] The specific working process and principle of the construction unit 110, the first generation unit 120, the processing unit 130 and the second generation unit 140 may refer to the relevant description of the aforementioned example.

[0155] It is understandable that the division of the above units is only a division of logical functions, and in actual implementation, they can be fully or partially integrated into one physical entity, or physically separated. In addition, the above units can be implemented in the form of a processor calling software.

[0156] The present embodiment also provides a computer system suitable for implementing the method for extracting key information from legal documents, wherein the computer system includes a central processing unit (CPU), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) or the program loaded from the storage part to the random access memory (RAM), such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in the RAM. The CPU, ROM and RAM are connected to each other via a bus. The input / output (I / O) interface is also connected to the bus.

[0157] The following components are connected to the I / O interface: an input part including a keyboard, a mouse, etc.; an output part including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage part including a hard disk, etc.; and a communication part including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication part performs communication processing via a network such as the Internet. The drive 310 is also connected to the I / O interface as needed. Removable media, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., are installed on the drive as needed so that the computer program read therefrom is installed into the storage part as needed.

[0158] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), various functions defined in the system of the present application are executed.

[0159] It should be noted that the computer-readable medium shown in the embodiments of the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system,

[0160] Device or component, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0161] In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable computer program. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0162] The flowchart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present application. Wherein, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0163] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware, and the units described may also be arranged in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0164] According to one aspect of the present application, a computer program product or a computer program is provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in the above-mentioned various optional implementations.

[0165] As another aspect, the present application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiment.

[0166] It should be noted that, although several modules or units of the equipment for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0167] Through the description of the above implementation methods, it is easy for those skilled in the art to understand that the example implementation methods described here can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the implementation methods of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the implementation methods of the present application.

[0168] Although the present specification is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. A method for quantitative prediction of mineral resources based on knowledge embedding, characterized in that: include: Construct a prospecting conceptual model based on the mineralization geological bodies, mineralization structural systems and mineralization characteristics associated with the target mining area; According to the prospecting concept model and the prospecting geological knowledge of the target mining area, a plurality of characteristic variables are obtained, and discrete and discontinuous characteristic variable sampling points of the target mining area are interpolated into continuous data by means of spatial interpolation to generate a prediction characteristic layer; Determining the weight of each of the characteristic variables; Generate a prediction model based on the weights of each of the characteristic variables and a pre-set ore deposit information characteristic depth network; Based on the prediction feature layer, a plurality of samples are generated, wherein one sample corresponds to one unit grid, a plurality of the unit grids constitute the prediction feature layer, and the samples include positive samples, negative samples and samples to be predicted; A plurality of the samples are input into the prediction model to obtain the mineral resource prediction result of the target mining area.

2. The method for quantitative prediction of mineral resources based on knowledge embedding according to claim 1 is characterized in that: The step of determining the weights of the characteristic variables comprises: Determine a first score and a second score for each characteristic variable; Compare the first feature vector with the highest first score with other feature variables, and determine a first score vector based on the relative importance between the first feature vector and other feature variables; Comparing the second feature vector with the lowest second score with other feature variables, and determining a second score vector based on the relative importance between the second feature vector and the other feature variables; According to each feature variable, a first scoring vector value of each feature variable in the first scoring vector, and a second scoring vector value of each feature variable in the second scoring vector, a BWM knowledge embedding model is used to determine a weight of each feature variable.

3. The method for quantitative prediction of mineral resources based on knowledge embedding according to claim 2 is characterized in that: The step of determining the first score and the second score of each feature variable comprises: Extract key information elements from each characteristic variable and determine the key information expression of each characteristic variable; Input the key information expression of each feature variable into the preset knowledge graph model to obtain multiple key nodes corresponding to each key information expression; Using the knowledge graph model, multiple key nodes of the key information expression are mapped to obtain multiple vector representations corresponding to each key information expression; The multiple vector representations corresponding to each key information expression and each key information expression are fused to form a fused variable corresponding to each feature variable; The fused variables corresponding to each feature variable are input into the pre-trained neural network, and under different confidence conditions, the first confidence and the second confidence of each fused variable are respectively obtained, and the first confidence is used as the first score, and the second confidence is used as the second score.

4. The method for quantitative prediction of mineral resources based on knowledge embedding according to claim 1 is characterized in that: The step of generating a prediction model according to the weights of each of the characteristic variables and the preset ore deposit information characteristic depth network comprises: The weight of the characteristic variable is input into the mineral deposit information characteristic depth network loss function to determine the loss value of the mineral deposit information characteristic depth network, wherein the loss value is determined based on the following formula: Based on the loss value, back propagation is performed using a gradient descent method to determine the parameter gradient of the mineral deposit information feature depth network; Based on the parameter gradient, using a parameter optimization algorithm, the weight parameters between different layers of the ore deposit information feature depth network are updated to generate the prediction model; Where N is the number of samples, C is the number of categories; y ij is a sample X i The true label of the j-th category in; is a sample X i The predicted probability of the jth category in w j is the weight.

5. The method for quantitative prediction of mineral resources based on knowledge embedding according to claim 1 is characterized in that: The sample is a positive sample, and the positive sample is used to characterize a mineral point; The step of generating a plurality of samples based on the predicted feature layer comprises at least one of the following: Gridding the predicted feature layer to generate a plurality of cells; generating a buffer zone with the mineral points in the target mining area as the center according to the spatial correlation of mineralization, and taking all samples in the buffer zone as initial samples, wherein the range of the buffer zone is determined based on the area of ​​the target mining area and the number of mineral points in the target mining area; Taking the cell where the initial sample point is located as a base point, any one cell from a plurality of neighboring cells is selected as an extended sample; performing difference processing on the extended sample and the initial sample to generate a plurality of the samples; The initial samples are labeled to generate initial samples with first label information and sample data with second label information are obtained; a preset supervision model is trained using the data with the first label information to obtain a learning model; the sample data with the second label information is input into the learning model to generate sample data with pseudo label information; the initial samples with the first label information and the sample data with the pseudo label information are input into the learning model to generate a plurality of the samples.

6. The method for quantitative prediction of mineral resources based on knowledge embedding according to claim 1, characterized in that: The sample is a negative sample, and the negative sample is used to characterize a non-mining point; The step of generating a plurality of samples based on the predicted feature layer comprises at least one of the following: Gridding the predicted feature layer to generate a plurality of cells, wherein one cell corresponds to one eigenvalue; calculating the statistics of each eigenvalue, and according to the statistics of each eigenvalue and the preset statistics, taking the samples corresponding to the cells outside the preset statistics as negative samples; Based on a nonlinear dimensionality reduction method, the high-dimensional data in the prediction feature layer is mapped to a low-dimensional space, the abnormal area is determined, and the samples corresponding to the cells in the abnormal area are taken as negative samples.

7. The method for quantitative prediction of mineral resources based on knowledge embedding according to claim 1 is characterized in that: The step of inputting the plurality of samples into the prediction model to obtain the mineral resource prediction result of the target mining area includes: Based on the positive sample, the negative sample and the sample to be predicted, as well as the first label corresponding to the positive sample, the second label corresponding to the negative sample and the third label corresponding to the sample to be predicted, and based on the prediction model, determine the mineralization probability of each unit grid; Based on the mineralization probability of each unit grid, the cumulative frequency method is used to determine the classification of the target area, and the cumulative result is used as the mineral resource prediction result of the target mining area, wherein the mineral resource prediction result of the target mining area includes the delineation of the predicted target area and the mineralization probability within the predicted target area.

8. The method for quantitative prediction of mineral resources based on knowledge embedding according to any one of claims 1 to 7, characterized in that: Before inputting the plurality of samples into the prediction model to obtain the mineral resource prediction result of the target mining area, the prediction method further includes: The prediction model is deconstructed and the parameters of the prediction model are adjusted using permutation importance, partial dependency plots, and model-independent local explanation algorithms.

9. A mineral resource quantitative prediction system based on knowledge embedding, characterized in that: include: A construction unit is configured to construct a prospecting conceptual model based on the mineralization geological bodies, mineralization structural systems and mineralization characteristic signs related to the target mining area; The first generating unit is configured to obtain a plurality of characteristic variables according to the prospecting conceptual model and the prospecting geological knowledge of the target mining area, and interpolate the discrete and discontinuous characteristic variable sampling points of the target mining area into continuous data by means of spatial interpolation to generate a prediction characteristic layer; A processing unit is configured to determine the weight of each of the characteristic variables; and based on the prediction characteristic layer, generate a plurality of samples, wherein one sample corresponds to one unit grid, a plurality of the unit grids constitute the prediction characteristic layer, and the samples include positive samples, negative samples and samples to be predicted; and input the plurality of the samples into a prediction model to obtain a mineral resource prediction result of the target mining area; The second generating unit is configured to generate the prediction model according to the weights of the respective characteristic variables and a preset mineral deposit information characteristic depth network.

10. An electronic device, characterized in that: include: At least one memory and at least one processor, the memory storing one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method for quantitative prediction of mineral resources based on knowledge embedding as described in any one of claims 1 to 8.

Images